Structure, Binding Energies, and IR-Spectral Fingerprinting of Formic

zero point energy correction1,2) suggests that the trans3 form is about 1000× ... 100 fs) pulse in the OH stretching ... compared with the cyclic dim...
0 downloads 0 Views 110KB Size
J. Chem. Theory Comput. 2008, 4, 533-541

533

Structure, Binding Energies, and IR-Spectral Fingerprinting of Formic Acid Dimers I˙ lhan Yavuz Physics Department, Marmara UniVersity, Go¨ztepe Kampus, Kadiko¨y 34772, Istanbul, Turkey Carl Trindle* Chemistry Department, The UniVersity of Virginia, CharlottesVille, Virginia 22904 Received June 27, 2007

Abstract: We describe equilibrium structures for a variety of species likely to be formed as intermediate species in the dimerization of formic acid to produce the stable C2h-symmetric doubly H-bonded dimer and perhaps produced as the vapor is irradiated. For several low-lying species the rearrangement pathways to the stable form are characterized at the MP2/6-311+G(d,p) level of theory, with optimized structures and vibrations computed with full counterpoise corrections for basis set superposition error. Estimates of vibrational frequencies with corrections for anharmonicity suggest that infrared transitions (CO stretches and OH out-of-plane motions) could signal the presence of species less stable than the C2h dimer, observable in irradiation studies of formic acid vapor.

Introduction Formic acid I exists primarily as the trans form (the H-CO-H angle ) 180) in the gas phase. The MP2/6-311+G(d,p) energy difference of 4.65 kcal/mol (4.40 kcal/mol after zero point energy correction1,2) suggests that the trans3 form is about 1000× more abundant than the cis form at room temperature.

Formic acid forms clusters in the gas phase. The structure and spectra of the C2h-symmetric dimer of formic acid II (below) has been thoroughly studied, both by computational modeling4 and by spectroscopic methods.5 Considerable emphasis has been placed on the proton exchange and the importance of tunneling in the process.6 * Corresponding author e-mail: [email protected].

10.1021/ct700161a CCC: $40.75

Shipman et al.7 have investigated the response of formic acid vapor to IR irradiation in the broad absorption associated with the OH stretch. The breadth has been rationalized by anharmonic coupling to lower-frequency modes, Fermi resonance with combinations of such modes, and (in the case of the symmetric dimer) Davydov coupling between the degenerate OH stretches. Low-temperature studies and temperature-dependent FTIR investigations have provided the basis for the study of intramolecular vibrational relaxation. The studies of Shipman et al. characterize vibrational relaxation of formic acid vapor near room temperature, subjected to an ultrashort (ca. 100 fs) pulse in the OH stretching region. Their observations suggest H-bond breaking with a characteristic time of about 20 ps and perhaps the existence of detectable amounts of a dimer other than the most stable C2h species. The feature associated with the possible new structure, which the authors term the “acyclic” dimer, is broad, centered at about 3230 cm-1. This may be compared with the cyclic dimer’s OH stretch at 3107 cm-1 and suggests weaker H-bonding in what might be a shortlived species. The broad feature evolves over a 100 to 200 ps duration. The authors consider the possibility that colli© 2008 American Chemical Society

Published on Web 02/06/2008

534 J. Chem. Theory Comput., Vol. 4, No. 3, 2008

Yavuz and Trindle

Chart 1. Structures of Species Discussed in the Text

Chart 2. Energies (kcal/mol) Relative to Two Trans Formic Acid Moleculesa

a Turi values from ref 8; QK values from Qian and Krim, ref 9; CVH values from Chocholousˇ ova´ , Vacek, and Hobza, ref 10; YT)this work (Yavuz and Trindle). Turi and QK have single-point CP corrections; CVH and YT used full counterpoise corrections in optimization. ZPE corrections are not included. TT refers to two isolated formic acid molecules in trans configuration; TC and CC have one and two cis species respectively.

sional cooling of the acyclic dimer(s) may account for the longer-time behavior but concluded that the growth in free OH absorption could not be rationalized in this way. Their preferred account is a dissociation of an acyclic dimer to monomers in the 100 to 200 ps time frame. Direct dissociation requires 14.8 kcal/mol (or about 5000 cm-1) according to photoacoustic measurement,8 so the 3000 cm-1 provided by IR irradiation must be augmented, perhaps by collision. The purpose of this investigation is to re-examine the energy demands for the formation of acyclic dimers and the further production of monomers for such intermediate species. We intend to identify species within the energy reach of the irradiation and to characterize their IR absorption spectra to provide a basis for more direct identification of the intermediate species. To this end we conduct computa-

tions employing correlation-corrected model chemistries (MP2 in extended basis sets), corrected by counterpoise compensation for Basis Set Superposition Errors and including estimates of anharmonicity and mode coupling. Modeling such small interactions as hydrogen bonds requires accurate methods. This includes a suitably large and flexible basis set, recognition of correlation corrections to energies and structures, and allowance for basis set superposition error (BSSE).9,10 According to Tzusuki et al.,3 the extrapolated basis set limit for the counterpoise (CP)corrected binding energy in CCSD(T) for the C2h-symmetric dimer of formic acid is 13.93 kcal/mol. Their estimate of the MP2 limit for the binding energy is 13.79 kcal/mol. We infer that these values are not ZPE-corrected. The landmark paper of Turi11 characterized this species and defined a standard notation for other equilibrium structures II-VIII of the dimers of trans-formic acid, using MP2 with basis sets up to D95++(d,p) and single-point counterpoise estimates of the basis set superposition error. Among the dimers of trans formic acid we find conventional linear )O...HO- and >O...HO- hydrogen bonds (II in the first case, III and IV in the second case), bent -OH...O< H-bonds which are probably slightly weaker (V and V′), shared H bonds, and much weaker CH...O< and CH...O) interactions (III, III′, IV, VI, VII, VIII). The binding energy can be approximately represented by the energies of the various types of H-bonds: the weak interaction CH...O on average is about 1 kcal/mol; OH...O is about 5-6 kcal/mol; and the distorted bond OH...O is about 3 kcal/mol. Qian and Krimm12 revisited these systems in their project to construct a suitable potential for molecular mechanics simulations. Their MP2/6-311++G(d,p) binding energies, counterpoise-corrected for BSSE at the equilibrium geometries, are in general agreement with Turi’s results. Chocholousˇova´, Vacek, and Hobza13 (CVH) have evaluated energies and structures of several of these species, with the MP2/aug-cc-pVDZ model chemistry including both

Formic Acid Dimers

J. Chem. Theory Comput., Vol. 4, No. 3, 2008 535

Table 1. New Equilibrium Structures and Energies Relative to Twice I-trans without and with Zero Point Energy Corrections (kcal/mol)

Chart 3. Counterpoise Corrections of Two Kinds, Single Point and Fulla

a Binding energy estimates for species II, III, IV, and V for various counterpoise corrections. Region A: left to right, MP2/6-31G(d) values with no CP correction, single-point CP, or optimized on a fully CPcorrected surface. Region B: left to right, analogous MP2/6311+G(d,p) values. All energies are in kcal/mol.

counterpoise-corrected optimization and single-point correction of the energy after conventional optimization. We have returned to these systems as the first step in a study of species that might be generated when formic acid vapor is irradiated by 3000 cm-1 photons. All our calculations employ Gaussian 03(W).14 In our best calculations our model chemistry is MP2/6-311+G(d,p) with full counterpoise corrections7 imposed throughout the optimization and frequency calculations. By “full counterpoise corrections” we mean that the method of Simon, Duran, and Dannenberg is employed.7 At every point in an optimization the energy and gradient E and ∇E are computed for the dimer FA-FA* (where FA and FA* are both formic acid molecules but may be differently disposed in space) in the joint basis BFA∪BFA*; for each monomer in the joint basis; and for each monomer in its own basis. Then the energy is written E ) E(FA + FA*;BFA ∪ BFA*) + E(FA;BFA ∪ BFA*) + E(FA*;BFA ∪ BFA*) E(FA*;BFA*) - E(FA;BFA) The counterpoise-corrected gradient and second derivative tensor require the derivatives of all four correction terms as well as the leading term. Simon, Duran, and Dannenberg report several cases, notably HF in water, where full CP produces significantly different structures and vibrational frequencies compared with single point CP correction.

Chart 2 displays various estimates of the energies of the nine species identified by Turi,8 relative to the dissociation products, two separated trans formic acid molecules. These values do not include zero-point energy corrections. We have interchanged species IV and V in CVH Table 1 (III and IV in their numbering), after reproducing their reported numbers for these species. It appears that values in the first column of CVH in Table 1 refer to binding energies obtained by conventional optimization followed by single-point counterpoise correction. Table S2 (Supporting Information) includes more detail of the results of Turi, Qian and Krimm, and Chocholousˇova´, Vacek, and Hobza as well as our own, including zero point energies and CP corrections. The key difference between our results and these values is the relative stability we find for species V which displays a -OH bond participating both as an acceptor and a donor in the sixatom ring stabilized by H bonds. Brinkmann, Tschumper, Yan, and Schaeffer15 have studied species I, II, and III, using a variety of basis sets and both MP2 and DFT correlation-corrected model chemistries. They estimate the binding energy of II to be about 15.9 kcal/mol (with MP2 with their TZ2P+dif basis) and of III to be about 9.5 kcal/mol. The counterpoise correction is about 2.4 kcal/ mol for II and 0.6 kcal/mol for III; this would shift the binding energies of II and III to 13.5 and 8.9 kcal/mol respectively. These values are apparently not corrected for zero-point vibrational energy. Our values seem to be consistent with Turi’s values, but our binding energy values are smaller for species II and III. This may arise either from details of the counterpoise corrections or effects of the difference in basis used; we locate minimum-energy structures and vibrational frequencies on the counterpoise-corrected potential, while Qian and Krimm and also Turi evaluate the CP correction using structures obtained by direct calculations. Chocholousˇova´, Vacek, and Hobza report results of both single-point CP corrections and counterpoise-corrected optimization. In general one expects smaller binding energies and lower interfragment frequencies for the counterpoise-corrected optimization compared to standard gradient optimization. Equilibrium structures found on the CP-corrected surfaces may be quite different in geometry and as well as energy relative to the analogous structures located on the uncorrected surfaces. One might wonder whether the relative energies found by a single CP correction are useful approximations to relative energies of structures found on the CP-corrected surface. Chart 3 bears on this question.

536 J. Chem. Theory Comput., Vol. 4, No. 3, 2008

Yavuz and Trindle

Table 2. Interatomic Distances for H-Bonding (Å) structural feature

I

-OH...O) (Turi) -OH...O) (Y-T) -OH...O) (CVH) -CH...O) (Turi) -CH...O) (Y-T) -CH...O) (CVH) structural feature

V

-OH...O< -OH...O) (Turi) -OH...O